Warning: These documents are under active
development and subject to change (version 2.1.0-beta).
The latest release documents are at:
https://purl.dataone.org/architecture
Environmental scientists commonly use the R Project software system for statistical analysis and modeling. The R framework is a flexible statistical computing environment with robust, extensive packages to perform a large variety of analysis and modeling tasks. Because R is open source and extensible, it has been widely adopted by the environmental science community, and it is easy to extend to provide new analysis capabilities.
The DataONE R Client Package is an R package that provides access to the DataONE services that are present at Coordinating Nodes and Member Nodes. The tool will allow an R user to easily load the d1r package (using CRAN) and then use functions in R to search the DataONE system, locate data of interest, load that data into the R environment, process it using R’s various tools, and then upload derived data and metadata back into DataONE.
By targeting R in the Investigator Toolkit, we enable a wide variety of scientists to harness data within the network of Member Nodes, and to reference the data sets within their R scripts using their unique identifiers from DataONE. This allows anyone to execute the R script from a paper or analysis, and the dataone library handles the details of finding and accessing the data needed for analysis or visualization.
A scientist with R on their system can easily install d1r from CRAN
A scientist can load a data object using its DataONE identifier from the DataONE system into R for further processing by other R functions. The data objects supported should minimally include:
- Data Tables in CSV and other delimited formats
- NetCDF files
- Raster images in various formats
A scientist can use d1r to search for data based on critical metadata, including title, creator, keywords, abstract, spatial location, temporal coverage, taxonomic coverage, and other relevant fields
A scientist can load all of the supported data objects into R by using the metadata found in a metadata search without directly knowing the identifiers for individual data objects
Given a data object loaded from DataONE into R, a scientist can display the science metadata associated with that object from within the R system (this might include a link to an external web URI that provides a nice human readable version of the metadata too)
A scientist can authenticate with the DataONE system in order to establish their identity to access restricted objects and services
A scientist can upload an R dataframe as a data set with associated science and system metadata for that object from within the R environment
Note
All of the classes and methods in R are inverted (ie, classes don’t contain methods per se, instead, methods are specialized to work on particular classes, and the class is generally passed in as the first parameter of the method call. So, all signatures below need to have an added parameter for this object to be passed in.
Note
This design is out of date, and instead follows the design in DataPackage. The R package mainly follows and uses the Java implementation, so does not need to reimplement this structure. Probably can delete this section, but leaving for now until DataPackage design is completed.
Methods